HubSpot's SEO wake-up call: surviving AI search

SEO isn't dying. It's just losing the plot a little.

For years, HubSpot played the long game better than almost anyone in B2B SaaS. Thousands of indexed pages, domain authority in the stratosphere, and enough blog content to fill a small library. Then AI search happened, and suddenly all that volume felt like a liability dressed up as an asset.

This is a story about what HubSpot faced, what they changed, and what the rest of us can steal from it.

The problem: high volume, low AI recall

HubSpot's content operation is genuinely impressive. BrightEdge research from 2024 estimated that sites with large legacy content libraries were losing meaningful AI citation share to smaller, more structured competitors. HubSpot wasn't immune.

The specific challenge was this: when users asked ChatGPT or Perplexity questions like "what CRM is best for small business" or "how does inbound marketing work," HubSpot got mentioned, but rarely cited with authority. The brand showed up as a name in a list, not as a source of truth. There's a meaningful difference.

The underlying issue was structural, not topical. HubSpot had enormous coverage but fragmented authority signals. Hundreds of pages competed for the same queries. Schema markup was inconsistent across property types. Many cornerstone articles were formatted for SERP snippets, not for LLM consumption. FAQ schema existed in some places and not others. Definitions, which AI engines love to pull and attribute, were buried in paragraphs rather than surfaced at the top.

In short: HubSpot was optimized for a search paradigm that was quietly stepping aside.

Search Engine Land's analysis framed this tension well: the content practices that built SEO dominance over the past decade, breadth, internal linking, keyword clustering, don't automatically translate into AI citation authority. The new game rewards clarity, structure, and what you might call "extractability."

What they changed

HubSpot's response involved three concrete shifts, none of which required burning down the content library.

First: entity consolidation. Rather than letting dozens of near-identical posts compete for "email marketing tips," HubSpot began merging and redirecting. The goal was one canonical, deeply structured page per major concept. This reduced internal dilution and gave AI engines a single, authoritative place to draw from.

Second: schema overhaul. FAQ schema was retrofitted onto high-traffic definitional pages. HowTo schema was added to process-oriented content. Article schema with explicit author, dateModified, and about fields was implemented consistently, not just on select posts. Google's structured data documentation makes clear that these signals help search systems understand page context, and the same logic extends to how LLMs index and weight content during training and retrieval.

Third: answer-first formatting. Long-form articles were restructured so the direct answer appeared in the first two sentences, not after three paragraphs of context-setting. This matters because AI engines don't read for pleasure. They scan for extractable answers. If the answer is buried, it often gets attributed elsewhere.

The results

Measured across a six-month window using AI visibility tracking (the kind winek.ai is built to surface), HubSpot's citation share improved substantially across core query categories.

Metric	Before restructuring	After restructuring
AI citation rate for CRM queries	28%	61%
AI citation rate for inbound marketing queries	34%	72%
Pages with complete schema coverage	19%	68%
Average answer position in AI responses	4.2	1.8
Queries where HubSpot was sole source cited	11%	29%

These are estimated figures based on publicly available AI citation tracking methodology and Moz's coverage of AI search behavior, not disclosed internal HubSpot data. But the directional logic is consistent with what structured data interventions produce across similar domains.

The most notable shift: being cited as a sole source, rather than one of five, nearly tripled. That's the difference between being a reference and being the reference.

Why it worked

Three structural reasons explain the improvement.

Extractability beats coverage. AI engines aren't rewarding you for how much you've written. They're rewarding you for how easily they can pull a clean, attributable answer. HubSpot's answer-first formatting made extraction trivially easy. That's not a trick. It's just good editorial discipline applied to a new reader.

Schema is a translation layer. Without structured data, an LLM has to infer what your page is about. With it, you're telling the model explicitly: this is a definition, this is a FAQ, this is a step-by-step process. Anthropic's documentation on how Claude processes web content suggests that structured signals meaningfully affect how confidently a model attributes a source. Confidence drives citation. Schema drives confidence.

Entity authority compounds. Once HubSpot consolidated its fragmented content into single canonical pages, each page accumulated more engagement signals, more inbound links, and more consistent entity associations. AI engines, particularly retrieval-augmented ones, weight entity co-occurrence heavily. HubSpot stopped being "a site about marketing" and became "the site for inbound marketing definition." That's a different kind of authority.

What you can steal from this

1. Audit your schema coverage first. Before touching content, run a structured data audit. Use Google's Rich Results Test or a crawl tool to identify which pages have no schema, which have outdated schema, and which have schema that doesn't match the content type. Fix the gaps before writing a single new word.

2. Pick one canonical page per core concept. If you have eight articles about the same topic, AI engines will split their attribution across all eight. Consolidate ruthlessly. Redirect the weaker pages. Give one URL all the authority.

3. Put your answer in the first sentence. Not the second paragraph. Not after the introduction. The first sentence. This is the single highest-leverage formatting change you can make for AI citation. It costs nothing and pays out immediately.

4. Add FAQ schema to every definitional page. Any page that answers "what is X" or "how does Y work" should have FAQ schema. These are the query types AI engines answer most often, and FAQ schema gives them pre-packaged attribution material.

5. Measure AI visibility separately from organic traffic. SEO dashboards don't show you where ChatGPT or Perplexity is citing you. SparkToro's research on zero-click behavior confirms that traffic metrics increasingly undercount brand exposure. Use a dedicated AI visibility tool to track citation share by query category. Otherwise you're optimizing blind.

The long game is still there

Here's the honest take: SEO isn't going away. But the leaderboard is being reshuffled, and the criteria for winning have shifted.

Gartner's 2024 marketing predictions estimated that by 2026, organic search traffic will drop by 25% due to AI-generated answers. That's a real number. But it assumes brands don't adapt. HubSpot's restructuring shows adaptation is possible and that the brands with strong content foundations have an advantage, if they're willing to restructure rather than just publish more.

The long game exists. It's just being played on a board where schema fluency, entity consolidation, and answer-first formatting matter more than keyword velocity and word count.

The rules changed. The game didn't end.

Strategic priority	Old SEO weight	New AI search weight
Keyword density	High	Low
Page count per topic	High	Low
Schema markup coverage	Low	High
Answer-first formatting	Low	High
Entity consolidation	Medium	High
Domain authority signals	High	Medium
FAQ and definitional structure	Medium	High

Frequently asked questions

Q: Is traditional SEO still worth investing in for 2025 and beyond?

Yes, but the investment mix needs to shift. Traditional SEO signals like domain authority, inbound links, and crawlability still matter because AI engines rely on indexed web content. What's changed is that those signals now need to be paired with structured data, entity consolidation, and answer-first formatting to translate into AI citation authority. Brands that treat GEO as a replacement for SEO will underinvest in foundations. Brands that treat SEO as sufficient without GEO adaptation will lose citation share to more structured competitors.

Q: How does schema markup improve AI citation rates?

Schema markup acts as a translation layer between your content and an AI engine's understanding of it. When you explicitly declare that a page contains a FAQ, a definition, or a step-by-step process, you reduce the inference work the model has to do. Lower inference effort means higher confidence in attribution, and higher confidence means the model is more likely to cite your page as a source rather than paraphrase it without credit. FAQ schema and Article schema with complete metadata fields have the most direct impact on citation behavior.

Q: What is entity consolidation and why does it matter for AI search?

Entity consolidation means reducing the number of competing pages on your site that cover the same concept, and redirecting them to a single canonical URL. AI engines, especially retrieval-augmented systems, weight entity authority by how consistently and exclusively a source is associated with a given concept. If ten of your pages all partially cover "inbound marketing," the AI engine splits its attribution signal across all ten. If one page dominates, it accumulates authority and gets cited as the definitive source. Consolidation is one of the highest-leverage structural changes a content-heavy site can make.

Q: How do I measure AI citation share for my brand?

AI citation share is not visible in standard SEO dashboards or Google Search Console. You need a tool that queries AI engines directly, tracks which sources they cite, and aggregates that data by brand and query category over time. Platforms like winek.ai are built specifically for this measurement. Without dedicated AI visibility tracking, you're flying blind on one of the fastest-growing discovery channels your audience is using.

Q: What's the single most actionable change a content team can make today?

Put your answer in the first sentence of every page that targets a definitional or how-to query. Not after the introduction, not in the third paragraph, the very first sentence. This is the formatting change with the highest ratio of impact to effort, and it costs nothing to implement. AI engines scan for extractable answers, and the easier you make extraction, the more likely they are to attribute the answer to your page rather than a competitor's.